You are on page 1of 10

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/6796893

Diagnostic Immunohistochemistry: What can


Go Wrong?

Article in Advances in Anatomic Pathology October 2006


DOI: 10.1097/01.pap.0000213041.39070.2f Source: PubMed

CITATIONS READS

42 170

2 authors, including:

Hadi Yaziji
Vitro Molecular Laboratories
97 PUBLICATIONS 3,697 CITATIONS

SEE PROFILE

All content following this page was uploaded by Hadi Yaziji on 02 June 2016.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
REVIEW ARTICLE

Diagnostic Immunohistochemistry: What can Go Wrong?


Hadi Yaziji, MD* and Todd Barry, MD, PhDw

pathology, dermatopathology, etc], but as a simple


Abstract: Although immunohistochemistry (IHC) has become technique.
an integral component of the surgical pathology laboratory and As shown in Table 1, pitfalls in the use of IHC can
practice, it remains a subspecialty in search of a definition. The occur at the technical, or design and interpretation phases
rapid expansion of this discipline is continually contributing to of work-up. Therefore, technologists, in addition to
the struggle of interested pathologists and technologists to pathologists, should be aware of these pitfalls. Ideally,
identify the proper working conditions of antibody reagents and the technologist and the pathologist should actually work
the right choice of detection systems and instrumentations. together in the validation, testing, design, and interpreta-
Training programs have not, for the most part, successfully tion of IHC assay. The benefits of such ideal working
incorporated IHC in their curricula. Thus, pitfalls in diagnostic relationship are immense.
IHC, in our opinion, are becoming more frequent than they used
to. In this review, we highlight and attempt to provide solutions
PITFALLS AT THE TECHNICAL LEVEL
for the most common pitfalls in diagnostic IHC as it relates to
the routine practice of surgical pathology.
Significant similarities exist between IHC and
cooking. In the kitchen, we select ingredients, the manner
Key Words: immunohistochemistry, pitfalls, immunophenotype, in which these ingredients are cooked (ie various heat
antibodies, antigens, detection systems, false-positive, false- sourcescharcoal, gas, electric, steamer, pressure coo-
negative, endogenous biotin, immunoreactivity, HIER, antigen ker), what type of sauce to add and the way we mix
retrieval everything together into a delicious dish. Similarly, in the
laboratory, proper fixation of the specimen, proper
(Adv Anat Pathol 2006;13:238246)
cutting of tissue sections, selection of the proper antibody
clone and buffer as well as the type of retrieval
(microwave, steam, water bath) are key to a successful
T he role of immunohistochemistry (IHC) and ancillary
technologies is critical to the practice of surgical
pathology. Historically, the integration of IHC into the
antibody immunostaining.

Specimen Fixation
daily surgical pathology practice has been slow and It thus should not come as a surprise that pitfalls in
unceremonious. In the late 1980s and early 1990s, most diagnostic IHC begin with the way we fix our specimens.
pathology practices were introduced to IHC with only a In fact, antigen degradation does occur even before tissue
limited panel of markers (ie, vimentin, S-100, k, l). The fixation. Fresh specimens that are subjected to prolonged
gradual and steadfast expansion of the antibody reper- drying before fixation can suffer significant loss of
toire has taken almost 2 decades, and now approaches antigenicity. With the increasing awareness of the harmful
several hundred antibodies that have been validated for effects of formaldehyde and increasing difficulty with its
clinical use (in vitro diagnostic) and are available for disposal, many new formaldehyde alternatives are now
diagnostic use. During most of this period, IHC has been commercially available and have replaced formalin in a
regarded primarily as an ancillary test, analogous to a small number of anatomic pathology laboratories. Most
special stain. of these fixatives are glyoxyl-based. The problem with
This steadfast expansion of IHC has been accom- many of these fixatives is that they may render the tissue
plished by considering IHC as an extension of special unsuitable for IHC methods. Examples include the drastic
stains. In a similar fashion, most residency training reduction of CD5 by B5 fixed tissue1 and the compro-
programs have not incorporated comprehensive and mised level of hormone receptors and Ki-67 in glyoxyl-
structured instruction of IHC into their curricula. To fixed tissue. The latter has equally disastrous effects on
date, many still do not view IHC as a distinct discipline in fluorescence in situ hybridization.2 Formaldehyde, despite
pathology [like gastro-intestinal (GI) pathology, hemato- its documented toxic effects, remains, to date, the favored
fixative of all specimens especially those requiring
hematopathology immunophenotyping.
The length of tissue fixation also has significant
From the *Ancillary Pathways, Miami, FL; and wPhenoPath Labora- effects on immunoreactivity. To paraphrase Dr Hector
tories, Seattle, WA.
Reprints: Hadi Yaziji, MD, Ancillary Pathways, LLC, 7533 SW 58th
Battifora, all tissues received in routine pathology
Avenue, Miami, FL 33143 (e-mail: yaziji@ancillarypath.com). laboratories are either under-fixed or over-fixed in
Copyright r 2006 by Lippincott Williams & Wilkins formalin, that is, none are optimally fixed for antigenic

238 Adv Anat Pathol ! Volume 13, Number 5, September 2006


Adv Anat Pathol ! Volume 13, Number 5, September 2006 Diagnostic Immunohistochemistry

601C can compromise some of the epitope binding sites.


TABLE 1. Pitfalls in IHC can Occur at 2 Main Levels
Microtomy sectioning of sections thinner than 3 mm or
Technical Pitfalls thicker than 7 mm will lead to suboptimal results. Thinner
Fixation
Processing sections will result in a very weak signal and thicker
Pre-treatment sections will lead to either loss of tissue section from the
Detection System glass slide or great difficulty in evaluating a thick section.
Professional pitfalls
Selection of panel Detection Systems
Sensitivity of panel
Choice of antibody type The choice and application of detection system are
Choice of antibody clone equally important. The current widely used detection
Interpretation of results system is the avidin biotin immunoperoxidase technique5
Interpretation of literature in which a primary unlabeled antibody is detected using a
secondary biotin-conjugated antibody from a different
species; detection is possible through a biotinylated
retention. However, in this postepitope retrieval era, over- peroxidase which is incorporated into a complex
fixation of tissue (by formalin) may not necessarily be as assisted by an avidin, which acts as a molecular glue
detrimental to antigenicity as previously thought. For binding the biotinylated secondary antibody to the
example, Poston et al3 have demonstrated the expression biotinylated peroxidase. A novel innovation in recent
of Hodgkin-related antigens in the original tissue years has been the polymer-based detection systems.6,7 As
obtained by Sir Thomas Hodgkin 150 years ago. It is with the avidin biotin immunoperoxidase technique, an
usually easier to overcome the problems of over-fixation unlabeled primary antibody is first employed, but in this
than under-fixation. The latter is often an unfortunate 2-step procedure, the second reagent is a dextran polymer
consequence of time pressures to minimize turnaround to which are bound large number of peroxidase molecules
time in pathology laboratories, despite the fact that we and secondary antibody molecules to bind to the
generally have no control over what time in the day unlabeled primary antibody. As will be noted below, this
tissues are obtained, and in what form the tissues are system has the advantage of being avidin-biotin-free,
received. Formalin takes more than a few hours to which eliminates the nonspecific biotin signal that can
adequately fix even small pieces of tissues,4 and it is not result in nonspecific staining in some tissues, particularly
uncommon for tissues arriving in an unfixed state in the liver and kidney. A number of polymer detection systems
surgical pathology laboratory at 5:00 PM to be cut and have recently become available.
placed in a tissue processor where the total immersion Although the avidin biotin and dextran-polymers
time in formalin will be less than 2 or 3 hours. In these immunohistochemical methods are currently state-of-
situations, because the next step in the tissue processor is the-art secondary reagents for diagnostic IHC, the next
invariably immersion into alcohol (which, in contrast to generation IHC technique involves a tyramine-based
formalin, fixes rapidly), late arriving tissues may be amplification methodology, also referred to as the
predominantly alcohol rather than formalin-fixed; worse, catalyzed reporter deposition or catalyzed system
they are partially alcohol and partially-formalin fixed, amplification method.8 These new detection systems can
which is particularly problematic for some antigens. The provide a highly sensitive method of detecting antigens
same can be said for cell blocks of body fluids, which may that may be present in low levels and thus, offer some
be processed in formalin once spun down and pelleted, advantages in this setting.
but the cells may have sat in an alcohol-based solution
before cell block processing. Similarly, under-fixed tissues Tissue Pretreatment
exhibit uneven IHC signal as edges may be better fixed Formalin fixation is thought to act on tissues
than the interior of tissue. through the formation of protein-protein cross-links,
The use of fixatives that have traditionally been presumably with the additional formation of coordinate
used for special tissues, such as B5 fixative for bonds for calcium ions.4,9 Alterations in the 3-dimen-
lymphoid tissues and Bouin fixative for gastrointestinal sional structure of proteins resulting from cross-linking
tract biopsies is a practice for which there is no longer change the 3-dimensional configuration of the native
good justification. Both of these fixatives manifest more antigen, altering antibody-antigen interactions. The
restricted antigen preservation than formalin, and the use introduction of epitope retrieval techniques in the early
of these fixatives not infrequently results in selective 1990s revolutionized the practice of diagnostic IHC by
antigenic loss that can be annoying at best and focusing attention on the undoing protein cross-
catastrophic at worst in the work-up of a given case. linking. Shi and colleagues,10 by placing tissues in a
microwave oven in a buffer, found a remarkable
Tissue Processing enhancement of immunostaining using 39/52 different
It is also critical to follow strict procedural steps antibodies tried, including antibodies to wide range of
before incubating the primary antibody on the tissue different antigens. Cattoretti et al11 also showed that new
section, to avoid detrimental staining results. For antibodies such as MIB-1, seemed to require microwave
instance, paraffin embedding at temperature in excess of pretreatment of the tissue for the antibody to work on

r 2006 Lippincott Williams & Wilkins 239


Yaziji and Barry Adv Anat Pathol ! Volume 13, Number 5, September 2006

fixed tissues. Several comprehensive studies of micro-


TABLE 2. Conventional and Preferred Panel for the Work-up
wave-based epitope retrieval have been reported, with of Undifferentiated Malignant Neoplasms
similar results.1214 Nonetheless, excessive tissue micro-
Diagnostic
waving can destroy antigenicity, and, reflecting gradients Category Conventional Panel Preferred Panel
in fixation, there can result gradients of immunoreactivity
Carcinoma Pan-cytokeratin Pan-cytokeratin
in the tissue. More commonly, areas of loose connective Lymphoma CD45 CD45+CD43
tissue and fat can display a very characteristic micro- Melanoma S-100 S-100 and either gp100 or
wave burn pattern. Techniques employing sources of MelanA
heat other than microwaves have been introduced. These Sarcoma Vimentin Collagen type IV and vimentin
include: autoclaving15; high temperature water bath;
pressure cooking16; and steam heating.17 Selection of
the pretreatment step may have a large impact upon the lymphomas, leukemias, and anaplastic plasmacytomas) and
final immunostained result.18 The combination of heat 2 melanocytic markers, S-100 and either gp100 (recognized
source, enzyme digestion, buffer type, and time of heating by clone HMB-45), or melanA, and collagen type IV and
can produce drastically different results. At our institu- vimentin as sarcoma markers (Table 2). It has been our
tions, when we acquire a new antibody, we experiment experience and others31 that the consistent expression of
with >3 types of pretreatment combinations along with collagen type IV around spindle cells in mesenchymal
>3 different dilutions until we achieve an optimal signal. tumors makes this marker useful in this differential
Pretreatment instructions included in antibodies package diagnostic setting.
inserts should be used as a guide, but they do not always
offer the best end result. The Design of Panel: Algorithm
It is very important to emphasize the algorithmic
PITFALLS AT THE DESIGN AND approach in the IHC work-up of any lesion, namely
INTERPRETATION LEVEL starting with a limited but highly sensitive panel, and
expanding with more specific markers depending on the
Selection of Panel results of the primary panel, as seen in Table 3 below.
The optimal goal of IHC techniques in the diagnosis A haphazard selection of markers is almost a guarantee
of surgical pathology is to maximize sensitivity without for missing the correct diagnosis (unless one empties the
compromising the specificity of the study results. To refrigerator and randomly applies >20 antibodies).
achieve a maximum sensitivity of target detection, it is This is particularly true if the pathologist has not
always a good practice to use more than one antibody for included a proper differential diagnosis for the IHC work-
a given antigenic target. For instance, if one suspects up of the tumor in question. For example, if one does not
neuroendocrine phenotype in a carcinoma, one could use consider the diagnosis of granulocytic sarcoma (chlor-
a combination of synaptophysin and chromogranin. It is oma) in the differential diagnosis, it practically does not
often (albeit not too common) that the lesion will be matter how many markers are applied unless such
positive with one or the other, but not both. In that markers as CD43, CD34, CD117 (c-kit), myeloperoxidase
respect, there should be no utility for neuron-specific (and others) are part of the panel.32 Other lesions that we
enolase in the modern IHC laboratory. This marker has have seen consistently been under-recognized are breast
proven its lack of specificity over the years. Another angiosarcomas and metastatic melanomas to solid organs
example of the importance of selecting antibodies in a and central nervous system (because of their epithelioid
panel relates to screening studies for undifferentiated morphology), as listed in Table 4. To paraphrase our
malignant neoplasms. The most common panel includes mentor, Dr G. Randolph Schrodt, in pathology, chance
cytokeratin, vimentin, S-100, and CD45, with the also favors the prepared mind.
perception that it will allow the pathologist to discrimi- A wise and safe approach would be to start the
nate (with high level of sensitivity) between carcinomas panel with the most general diagnostic categories in mind
(keratin positive), sarcomas (vimentin positive), lymphomas and then to narrow the scope of the IHC panel to more
(CD45 positive), and melanomas (S-100 positive). In reality, specific antibodies based on the results of the initial panel.
vimentin is positive on many carcinomas (thyroid, renal cell,
any carcinoma with spindle cell features),1922 mesothelio-
mas, and lymphomas.23 It is negative on some sarcomas (ie, TABLE 3. Algorithm Approach in the Design of IHC Panels
alveolar soft part). Cytokeratin should be absent on adrenal Algorithm Component Description
cortical carcinomas24 and positive on many sarcomas
(especially leiomyosarcomas)2529 and melanomas.24,30 S- Step 1: maximum Start with a panel that offers the most
sensitivity sensitive and comprehensive results
100 is undoubtedly absent on a subset of melanomas (about Step 2: maximum Apply more specific markers to confirm
2% to 10%) and positive on many carcinomas (particularly specificity the results obtained in step 1
thyroid, salivary gland, and renal cell). CD45 is also absent Step 3: final diagnosis May not be necessary if step 2 work-up
on some lymphomas (classic Hodgkin) and leukemias. offered definitive results
Usually accomplished with 1 or 2
A more realistic panel should include cytokeratin markers
(for carcinomas), CD43+CD45 (which will cover most

240 r 2006 Lippincott Williams & Wilkins


Adv Anat Pathol ! Volume 13, Number 5, September 2006 Diagnostic Immunohistochemistry

low sensitivity of calretinin in detecting mesothelio-


TABLE 4. Frequently Encountered Misclassified or Under-
recognized Lesions in our Practices ma.35 The authors used a calretinin clone that should
have been discontinued and neglected to use the more
Organ Entity Mistaken asy
widely used clone. Unfortunately, the pathology literature
Lymphoid Granulocytic sarcoma PD Carcinoma in the past 15 years is cluttered with such false claims of
(nodes, BM, etc)
CD30-negative Hodgkin Anaplastic large cell
either superior or poor antigens, whereas in reality it is
Breast Angiosarcoma Carcinoma the choice of a given antibody clone (and not the antigen)
Sclerosing adenosis Invasive carcinoma that is the source of such claim. Selection of the
Any organ Melanoma Carcinoma appropriate clone can have a large impact on the
detection of a given antigen.
In some situations, there is yet to exist a reliable
For example, if a pathologist faced with a pleural-based antibody clone. A common pitfall in the selection of such
epithelioid tumor designs a panel of markers directed that markers includes antibodies to CA-125. This is arguably a
address the differential diagnosis of carcinoma of useful serum marker that guides the oncologists to search
unknown primary site, he may miss the diagnosis of for and detect ovarian tumors; however, the tissue-based
mesothelioma altogether. Thus, the initial panel chosen equivalent has a very compromised specificity. In our
should be broad in scope, first excluding mesothelioma experience, it has been positive on GI and breast tumors.
and establishing the diagnosis of carcinoma and then We do not offer CA-125 antibody test in our respective
proceeding to a more focused work-up of a lung primary laboratories. An excellent alternative is WT-1 antibody
versus metastatic carcinoma to the pleura. for the detection of ovarian serous tumors.36
Using the principle of maximum sensitivity and
specificity, one should always attempt to fine tune Specific and Nonspecific Immunoprofiles
existing panels to address varying differential diagnosis Sometimes, it is our lack of knowledge that certain
considerations, not to mention that such panels must be lesions exhibit specific and reproducible immunoprofiles
designed in the context of morphologic and clinical that hinders our ability to use IHC to confirm (or refute)
features of a given tumor. On occasion, the clinical the diagnosis of these lesions. For instance, the combina-
information offers the most sensitive and specific data to tion of CK+, MOC31 " , CD10+ (in a bile canalicular
help guide selection of an antibody. A 67-year-old woman pattern), and CD34+ (in a sinusoidal pattern) profile is
with a history of breast cancer and a recently discovered virtually diagnostic of hepatocellular carcinoma. Table 5
tumor in the lung most likely has either metastatic breast contains a list of entities with specific immunophenotypes.
cancer or primary lung cancer, and there is no need to Conversely, expression of the cdx-2 GI transcription
include GI and ovarian markers in the primary panel, factor is not at all specific for GI tumors. As we and
which could be limited to mammary and lung markers
(TTF-1, surfactant, ER, PR, GCDFP-15, or mammaglo-
bin). If the results of the primary panels are not definitive,
TABLE 5. Examples of Pathologic Entities With Unique
then one can apply a secondary panel of markers that Diagnostic Immunophenotypes
should cover other solid organs.
Pathologic Entity Unique Immunophenotype
Importantly, there is always a continual evolution
of antibodies over time. Markers that were once Hepatocellular CD34 in a sinusoidal pattern, collagen
considered state-of-the-art are eventually replaced by carcinoma type IV around tumor nests
Meningioma Progesterone receptor, EMA
markers that prove to be more sensitive and specific. For Hemangioblastoma S-100, EGFR
instance, a panel of antibodies used in the differential Barrett esophagus CDX-2 on intestinalized cells
diagnosis of adenocarcinoma versus mesothelioma once Hydatidiform mole P57 (for complete moles), FISH (for triploid
included antibodies such as vimentin, carcinoembryonic partial moles)
Hairy cell leukemia CD20, CD25, TRAcP, cyclin D1
antigen, epithelial membrane antigen, and B72.3. How- Fibromatosis B-catenin (nuclear signal), smooth muscle
ever, more sensitive and specific markers now can be used actin (not desmin)
in this differential. Positive carcinoma markers should Glomus tumor Smooth muscle actin (not desmin),
include MOC-31, BG8, and BerEP4 (as epithelial synaptophysin
glycoproteins) while positive mesothelioma markers Endometrioid Vimentin, B-catenin (nuclear)
carcinoma
should include calretinin, WT-1, cytokeratin 5, and PML SV40, p53 (invariably positive)
mesothelin.33 It is usually sufficient to establish the Monophasic synovial Keratin, CD99, b catenin
diagnosis with only 2 epithelial and 2 mesothelial sarcoma
markers. The recently described lymphatic endothelial PEComa Smooth muscle actin, gp100 (HMB-45)
Kaposi sarcoma HHV8
marker podoplanin seems to be highly sensitive and Follicular dendritic cell CD21, CD35, CD23
specific mesothelial marker.34 tumor
Inflammatory MFB Smooth muscle actin (filamentous pattern),
The Choice of Antibody Clone tumor EBV, ALK-1
Not all antibody clones are created equal. A typical Merkel cell tumor Ck20-positive (dotlike), synaptophysin-
positive, TTF-negative
example is a study published in 2001 that described the

r 2006 Lippincott Williams & Wilkins 241


Yaziji and Barry Adv Anat Pathol ! Volume 13, Number 5, September 2006

others documented,37 cdx-2 is expected to be positive on there is endogenous biotin present in many tissues
ovarian mucinous tumors and bladder adenocarcinomas (particularly liver and kidney), and also many tumors.
because of the GI phenotype of these tumors. Not to The latter tissues, in particular, can yield false-positive
forget the primary mucinous lung carcinomas with GI signals with avidin biotin-based techniques, as amplifica-
(TTF " , surfactant " , cdx2+) phenotype.38 These are tion of the endogenous biotin can be just as effective in
usually adenocarcinomas with partial bronchioloalveo- yielding a brown (or black, or red, etc.) signal as can the
lar/mucinous features. Also, as mentioned above, mela- binding of the primary antibody to its target antigen.
nomas and leiomyosarcomas can often express Usually, endogenous biotin signal can be distinguished by
cytokeratin.39,40 This observation emphasizes the danger its fine, granular cytoplasmic pattern; while this can help
of using a single marker to attempt to establish a if the primary antibody is expected to yield a nuclear
diagnosis. signal, if the antibody being employed is expected to yield
Another common pitfall, we encounter in our a granular cytoplasmic pattern (eg, antibodies to chro-
practice is the reflexive designation of CD30+/CD15 " mogranin A), this can pose a serious problem. In addition
tumor as anaplastic large cell lymphoma. It is not of using of biotin block steps to avoid this problem, we
uncommon to receive a request from a colleague to strongly advise the use of alternative, avidin and biotin-
perform p80/ALK-1 studies to confirm anaplastic large free detection systems such as the polymer-based
cell lymphoma on tumors with the above results. With reagents. We favor the latter.
minimal work-up (absence of CD3, CD43), one can The issue of false-positive immunostaining cannot
suspect the possibility of CD15-negative Hodgkin lym- be dismissed as only a function of inappropriately high
phoma, not to mention the more common diagnosis of antibody concentrations or other factors (eg, biotin)
CD30-positive diffuse large B-cell lymphoma, which can unrelated to the primary antibody. Antibodies may, in
be established by showing expression of CD79a (and/or fact, stick to tissue by mechanisms other than antigen-
CD20), and coexpression of CD45 (along with absence of antibody binding, and these can be some of the most
CD3, CD43, and ALK1). Similarly, one could easily difficult artifacts to identify. Examples of undesired
make a misdiagnosis of grade I follicular lymphoma specific signal include the apparent specificity of many
simply on the basis of BCL-2 expression on the lymphoid different antibodies for pancreatic islet cells, the stick-
follicles. Resting or primary follicles by definition do not ing of antiprogesterone receptor antibodies to mucin
have germinal centers and thus are composed primarily of globules, and the nuclear signals sometimes seen with the
BCL-2+ mantle zone B cells, which can impart the L26 anti-CD20 antibody. Help in distinguishing true,
impression of a BCL-2+ follicular neoplasm. However, desired immunostaining from true, undesired immuno-
these cells show prominent IgD positivity, characteristic staining is often found by noting the intracellular
of resting follicles that are not typical of follicular localization of the signal.
lymphoma. Antibodies define antigens, which have, in general, a
Specific knowledge of the right expression (staining) well-characterized locus and pattern of expected signal.
pattern of a given marker is also crucial. As noted above, There are antigens which are exclusively nuclear (eg, TdT,
the most common (and costly) error in interpretation of TTF-1, estrogen receptor), antigens exclusively cytoplas-
markers that we have observed over the years is false- mic, (eg, cytokeratins, thyroglobulin, surfactant ApoA),
positive interpretation of endogenous biotin. A review of antigens with combined nuclear and cytoplasmic localiza-
the etiologies of false-positive signal is in order: Gen- tions (eg, S-100 protein, calretinin), antigens localized to
erally, when all cells and tissues appearing positive on a the cell membrane (eg, CD45, HER-2/neu), antigens with
given slide is caused by excess antibody concentration, an complex distributions (eg, the membranous plus Golgi
example of nonspecific signal. A general rule of thumb distribution of CD15 and CD30 in Hodgkin cells), and
is that antibodies should be used on tissue sections at a antigens with exclusively extracellular localization
protein concentration of approximately 1 mg/mL (or less). (eg, type IV collagen).
However, some can tolerate higher concentrations, and Different antibodies often yield characteristic pat-
others will start nonspecifically sticking to tissue at even terns in addition to characteristic localizations within
2 or 4 times the recommended concentration. Some cells. For example, antibodies to S-100 produce a
antibodies (eg, HMB-45) will appear to react with all cells homogenous signal; antibodies to cytokeratins can
in tissues as a function of inappropriate fixation (eg, B5 produce a filamentous signal; and antibodies to chromo-
fixed tissue). As has been noted, the distinction of granin A generally produce a granular signal. (These
immunostaining (which can be artefactual) and expres- patterns in part reflect the subcellular distribution of the
sion is an important one.41 False-positive immunostain- antigen in question, eg, the localization of chromogranin
ing can be caused by the use of antibodies at A to discrete dense core granules.)
inappropriately high concentrations,42 potentially altered Different antibodies also can yield characteristic
specificities following the use of epitope retrieval techni- patterns across tissues or tumors. For example, antibodies
ques,43 and the cross-reactivity of anticytokeratin anti- to nuclear transcription factors such as TTF-1 and CDX-
bodies with other proteins.44 Another very common 2, in lung carcinomas and colorectal adenocarcinomas,
reason for false-positive immunostaining is the adverse respectively, when positive on a given tumor, are positive
effects of avidin biotin-based IHC detection systems: in most tumor cells, whereas markers to proteins of

242 r 2006 Lippincott Williams & Wilkins


Adv Anat Pathol ! Volume 13, Number 5, September 2006 Diagnostic Immunohistochemistry

terminally differentiated cells such as the gross cystic specimens. For instance, cyclin D1 and CD30 can be
disease fluid protein-15 of breast cancers or uroplakin of severely compromised in B5-fixed tissue.47 Decalcification
urothelial carcinomas, are generally positive in only a has long been known for having negative impact on the
small fraction of the tumor cell population. tissue antigenicity.48,49 Not to mention the same issues
All these factors (the specificity of the antibody, its related to general diagnostic IHC would also apply to this
subcellular distribution, its pattern of reactivity within a topic, including suboptimal epitope retrieval and impro-
given cells and its pattern of reactivity within a popula- perly titered antibodies. What is also unique to lympho-
tion of cells, together with known possible undesired proliferative disorders is the frequent occurrence of target
specific immunostaining tendencies), constitute an anti- antigens expressed at low levels. Examples include low
body personality profile, which is something generally levels of CD2050 and ZAP-7051 in CLL/SLL. Another
learned with experience and often not provided in the common pitfall especially in the work-up of previously
package insert. Part of the antibody personality profile, treated B-cell lymphomas with anti-CD20 Rituxan is the
however, is a multidimensional matrix of fixation factors, virtual elimination of CD20 signal as a result of
antibody concentration, tissue pretreatment, and detec- treatment.52 CD79a would be one good alternative as a
tion systems, which are the optimal operating space for B-cell marker.
a given antibody. If you are inadvertently operating Artifactual staining can sometimes be a source of
outside of this optimal operating space, that is, in a region interpretation problems, such as the deposition of
of inappropriately high antibody concentration or in- mercury in B5-fixed tissue.47,53 As mentioned above,
appropriately fixed tissue, a false-positive signal may many are familiar with the specific and reproducible
result. nucleolar CD20 signal on non-neoplastic cells.

Degree of Expression is Contextual Interpretation Aspects


Sometimes expression of a marker on a subset of Table 6 below summarizes the different patterns of
tumor cells is as insignificant as no expression at all and antigen localizations in lymphoid lesions. As we discussed
sometimes these findings are just the tip of the iceberg in a above, lack of knowledge of subcellular localization of a
specimen that may be poorly fixed. For instance, given antigen may lead to problems in correctly inter-
expression of any of the intermediate filament cytoke- preting results.
ratins on a few tumor cells in a given carcinoma can, with Perhaps, the least known pitfall in diagnostic IHC
few exceptions (ie, small cell carcinomas) be dismissed as of hematolymphoid disorders is the infidelity of tumor-
negative. A metastatic carcinoma to the liver with a specific markers. Table 7 highlights a list of potential
uniform cytokeratin 20 signal and only rare cells positive anomalous expression of various markers that have been
for cytokeratin 7 signal should not be labeled as well documented in the literature. For example, CD20
cytokeratin7+/20+ tumor. Conversely, if a metastatic expression has been reported in a subset of peripheral T-
carcinoma that shows coexpression of cytokeratins 7 and cell lymphomas.54 Also, BCL-2 is negative in a subset of
20 on >20% of tumor cells also exhibits a weak and follicular lymphomas, especially those of higher grade.55
focal cdx-2 signal, the latter should not be dismissed as Another common problem in diagnostic IHC of
negative. Low levels of cdx-2 are not uncommon among lymphomas is the difficulty in demonstrating light chain
pancreatic and gastric carcinomas. The only GI tumors restriction. It is usually easier to highlight either k or l
that almost always express high levels of cdx-2 are those light chain restriction in neoplasms with plasmacytoid
of colorectal origin.37 differentiation.56 However, there are occasional circum-
Scoring of a signal as positive or negative is not only stances in which light chain restriction may not be
based on the statistical probability of significance of such sufficient for the diagnosis of lymphoma. This pheno-
signal, but sometimes on the clinical meaning of various menon has been described in Castleman disease,57 and in
levels of expression. In the field of prognostic factors, marginal zone hyperplasia.58
expression of estrogen receptor on >1% of a breast Finally, the use of ancillary molecular tools is
cancer cells has been shown to be associated with clinical always helpful in assessing B or T-cell clonality (by
response to tamoxifen.45 On the other hand, even in performing immunoglobulin heavy chain and T-cell
tumors with high levels of estrogen receptor expression, receptor gene rearrangement, respectively). Furthermore,
absence of progesterone receptor expression has been
shown to predict a poor outcome, hence the significance
of testing for both receptors at the same time.46
TABLE 6. Localization of Various Hematolymphoid Markers
in Diagnostic IHC
SPECIAL SITUATIONS: IHC IN
Localization Examples
HEMATOPATHOLOGY
Nuclear BCL-6, PAX-5, OCT-2, BOB-1
Technical Aspects Nuclear+cytoplasmic BCL-10
Perhaps the most important technical problems Cytoplasmic BCL-2
Membranous CD20, CD3, CD5, CD23, CD43, CD79a, etc
related to IHC in hematopathology are the use of Membranous+Golgi CD15, CD30
nonformalin fixatives on bone marrow and lymph node

r 2006 Lippincott Williams & Wilkins 243


Yaziji and Barry Adv Anat Pathol ! Volume 13, Number 5, September 2006

website (Immunoquery) are to be commended for their


TABLE 7. Tumor-specific Markers and Their Infidelity
Examples efforts, we do not recommend using it as the main source
of information. This is because the site combines the
Marker(s) Example References
literature on any given subject together as if all the studies
CD20 Expression in PTCLs 52 performed on that subject are of equal quality. Moreover,
CD79a Expression in T-ALL 54
TdT Expression in AML 54
different Ab clones may be used in different studies and
CD30 and CD30+/CD15 " in 10-15% of classic 55 the definition of positivity may be defined differently in
CD15 Hodgkin various studies, making comparison of the studies next to
CD30+/CD15 " in ALCL 56, 57 impossible. The reality is that the data gleaned from this
CD30 and CD15 expression in diffuse 58, 59 website may not be very helpful at best and misleading in
large BCL
CD30 and CD15 expression in PTCL 51, 59 some instances.
BCL-2 Expressed in mantle zones of primary
follicles
Negative in a subset of follicular 53
lymphomas HOW TO MINIMIZE THE ERRORS IN IHC
B-cell markers Expressed in classic HL 60
PAX-5 can be seen in NE tumors At the technical level, no antibody should be
CD138 Can be expressed in epithelial tumors acquired without the basic knowledge of its performance
characteristics and the expected expression (staining)
ALCL indicates anaplastic large cell lymphoma.
pattern. If possible, the avidin-biotin detection system
should be replaced by a polymer-based one to minimize
the sources of false-positive endogenous biotin signal.
identification of specific translocations either by PCR or More importantly, pathologists should be working closely
more recently by fluorescence in situ hybridization has with the technologists to educate one another about the
also shown its utility in the routine of hematopathology various issues discussed in this material. The chance for
diagnosis of hematolymphoid neoplasms. things going wrong in the IHC laboratory is real and not
a perceived one, and only a skilled technologist can
LIMITATIONS OF IHC function as the gatekeeper with the ability to identify and
troubleshoot various problems in the laboratory.
Table 8 shows a list of diagnostic entities where IHC
It is very difficult, even for an experienced patho-
should play a limited role. As an example, one of the
logist, to keep up with the literature on new and existing
common requests that we receive is to perform IHC
markers, making the field of IHC one of the fastest
studies to distinguish between primary squamous cell
growing and more exciting subspecialties to study in
carcinoma of the lung and metastatic squamous cell
pathology. Therefore, pathology groups should have at
carcinoma from the head and neck. These 2 types of
least one member with the expertise and interest to be
tumors are immunophenotypically identical. The only
responsible for the day-to-day laboratory functions and
subtypes of squamous carcinoma that show some
the interpretation and design of IHC panels.
discriminating features are nasopharyngeal carcinoma
The algorithmic approach of highly sensitive
(positive for Epstein-Barr virus genome59,60) and cervical
screening panels followed by highly specific diagnos-
squamous carcinoma (sometimes positive for estrogen
tic panels should be adopted. This practice is safe and
receptor). Similarly, both skin appendage tumors and
constitutes good medicine. After all, the main goal of
squamous cell carcinomas of the skin can show similar,
performing IHC studies is to strengthen and extend our
if not identical immunophenotypes.
hematoxylin and eosin (H&E) diagnostic power. The IHC
literature should always be interpreted with caution and
LITERATURE SEARCH at least some degree of skepticism. Moreover, it is
In addition to Medline, Immunoquery has become important to gain experience with particular antibodies
an increasingly popular Internet website and the top and continually compare your experience with that of the
search engine of choice for many pathologists seeking established literature.
specific information on IHC. Although the editors of that Residency training programs that have not fully
incorporated IHC in their curricula, should do so with the
help of experienced pathologists. The highly expensive
TABLE 8. Examples of Diagnostic Entities Where IHC Still IHC reagents and their availability make it practically
has a Limited Role impossible for any pathology to be self-taught without the
proper environment.
Distinction between squamous carcinomas of various organs
Prediction of stromal invasion in cervical and vaginal squamous In summary, the same guiding common sense
intraepithelial lesions principles that a great pathologist applies to his/her
Distinction between DH, ADH, and DCIS in the breast approach to H&E diagnoses can be applied to the
Assessment of stromal invasion in cutaneous squamous carcinomas practice of IHC. Those of us who make fewer errors in
Distinction between skin appendage tumors and primary squamous or
basal cell carcinomas
their conventional H&E practice will likely be making
fewer errors when dealing with IHC.

244 r 2006 Lippincott Williams & Wilkins


Adv Anat Pathol ! Volume 13, Number 5, September 2006 Diagnostic Immunohistochemistry

REFERENCES 23. Gustmann C, Altmannsberger M, Osborn M, et al. Cytokeratin


1. Chen CC, Raikow RB, Sonmez-Alpan E, et al. Classification expression and vimentin content in large cell anaplastic lymphomas
of small B-cell lymphoid neoplasms using a paraffin section and other non-Hodgkins lymphomas. Am J Pathol. 1991;138:
immunohistochemical panel. Appl Immunohistochem Mol Morphol. 14131422.
2000;8:111. 24. Pinkus GS, Etheridge CL, OConnor EM. Are keratin proteins a
2. Tubbs RR, Hsi ED, Hicks D, et al. Molecular pathology testing of better tumor marker than epithelial membrane antigen? A compara-
tissues fixed in prefer solution. Am J Surg Pathol. 2004;28:417419. tive immunohistochemical study of various paraffin-embedded
3. Poston RN. A new look at the original cases of Hodgkins disease. neoplasms using monoclonal and polyclonal antibodies. Am J Clin
Cancer Treat Rev. 1999;25:151155. Pathol. 1986;85:269277.
4. Fox CH, Johnson FB, Whiting J, et al. Formaldehyde fixation. 25. Eusebi V, Carcangiu ML, Dina R, et al. Keratin-positive epithelioid
J Histochem Cytochem. 1985;33:845853. angiosarcoma of thyroid. A report of four cases. Am J Surg Pathol.
5. Hsu SM, Raine L, Fanger H. Use of avidin-biotin-peroxidase 1990;14:737747.
complex (ABC) in immunoperoxidase techniques: a comparison 26. Miettinen M. Keratin subsets in spindle cell sarcomas. Keratins are
between ABC and unlabeled antibody (PAP) procedures. widespread but synovial sarcoma contains a distinctive keratin
J Histochem Cytochem. 1981;29:577580. polypeptide pattern and desmoplakins. Am J Pathol. 1991;138:
6. Sabattini E, Bisgaard K, Ascani S, et al. The EnVision++ system: 505513.
a new immunohistochemical method for diagnostics and research. 27. Corson JM. Keratin protein immunohistochemistry in surgical
Critical comparison with the APAAP, ChemMate, CSA, LABC, pathology practice. Pathol Annu. 1986;21:4781.
and SABC techniques. J Clin Pathol. 1998;51:506511. 28. Miettinen M, Virtanen I, Damjanov H. Coexpression of keratin and
7. Kammerer U, Kapp M, Gassel AM, et al. A new rapid vimentin in epithelioid sarcoma. Am J Surg Pathol. 1985;9:460463.
immunohistochemical staining technique using the EnVision anti- 29. Miettinen M, Lehto VP, Virtanen I. Keratin in the epithelial-like
body complex. J Histochem Cytochem. 2001;49:623630. cells of classical biphasic synovial sarcoma. Virchows Arch B Cell
8. Adams JC. Biotin amplification of biotin and horseradish per- Pathol Incl Mol Pathol. 1982;40:157161.
oxidase signals in histochemical stains. J Histochem Cytochem. 30. Miettinen M, Franssila K. Immunohistochemical spectrum of
1992;40:14571463. malignant melanoma. The common presence of keratins. Lab
9. Werner M, Chott A, Fabiano A, et al. Effect of formalin tissue Invest. 1989;61:623628.
fixation and processing on immunohistochemistry. Am J Surg 31. Leong AS-Y, Vinyuvat S, Suthipintawong C, et al. Patterns of basal
Pathol. 2000;24:10161019. lamina immunostaining in soft-tissue and bony tumors. Appl
10. Shi SR, Key ME, Kalra KL. Antigen retrieval in formalin-fixed, Immunohistochem. 1997;5:17.
paraffin-embedded tissues: an enhancement method for immuno- 32. Menasce LP, Banerjee SS, Beckett E, et al. Extra-medullary myeloid
histochemical staining based on microwave oven heating of tissue tumour (granulocytic sarcoma) is often misdiagnosed: a study of 26
sections. J Histochem Cytochem. 1991;39:741748. cases. Histopathology. 1999;34:391398.
11. Cattoretti G, Becker MH, Key G, et al. Monoclonal antibodies 33. Trupiano JK, Geisinger KR, Willingham MC, et al. Diffuse
against recombinant parts of the Ki-67 antigen (MIB 1 and MIB 3) malignant mesothelioma of the peritoneum and pleura, analysis of
detect proliferating cells in microwave-processed formalin-fixed markers. Mod Pathol. 2004;17:476481.
paraffin sections. J Pathol. 1992;168:357363. 34. Ordonez NG. Podoplanin: a novel diagnostic immunohistochemical
12. Gown AM, de Wever N, Battifora H. Microwave-based antigenic marker. Adv Anat Pathol. 2006;13:8388.
unmasking: a revolutionary new technique for routine immuno- 35. Roberts F, Harper CM, Downie I, et al. Immunohistochemical
histochemistry. Appl Immunohistochem. 1993;1:256266. analysis still has a limited role in the diagnosis of malignant
13. Cattoretti G, Pileri S, Parravicini C, et al. Antigen unmasking on mesothelioma. A study of thirteen antibodies. Am J Clin Pathol.
formalin-fixed, paraffin-embedded tissue sections [see comments]. 2001;116:253262.
J Pathol. 1993;171:8398. 36. Hwang H, Quenneville L, Yaziji H, et al. Wilms tumor gene
14. Cuevas EC, Bateman AC, Wilkins BS, et al. Microwave antigen product: sensitive and contextually specific marker of serous
retrieval in immunocytochemistry: a study of 80 antibodies. J Clin carcinomas of ovarian surface epithelial origin. Appl Immuno-
Pathol. 1994;47:448452. histochem Mol Morphol. 2004;12:122126.
15. Bankfalvi A, Navabi H, Bier B, et al. Wet autoclave pretreatment 37. Werling RW, Yaziji H, Bacchi CE, et al. CDX2, a highly sensitive
for antigen retrieval in diagnostic immunohistochemistry. J Pathol. and specific marker of adenocarcinomas of intestinal origin: an
1994;174:223228. immunohistochemical survey of 476 primary and metastatic
16. Norton AJ, Jordan S, Yeomans P. Brief, high-temperature heat carcinomas. Am J Surg Pathol. 2003;27:303310.
denaturation (pressure cooking): a simple and effective method of 38. Goldstein NS, Thomas M. Mucinous and nonmucinous bronchio-
antigen retrieval for routinely processed tissues. J Pathol. 1994; loalveolar adenocarcinomas have distinct staining patterns with
173:371379. thyroid transcription factor and cytokeratin 20 antibodies. Am
17. Taylor CR, Shi SR, Chen C, et al. Comparative study of antigen J Clin Pathol. 2001;116:319325.
retrieval heating methods: microwave, microwave and pressure 39. Zarbo RJ, Gown AM, Nagle RB, et al. Anomalous cytokeratin
cooker, autoclave, and steamer. Biotech Histochem. 1996;71:263270. expression in malignant melanoma: one- and two-dimensional
18. Shi SR, Cote RJ, Taylor CR. Antigen retrieval immunohistochem- western blot analysis and immunohistochemical survey of 100
istry: past, present, and future. J Histochem Cytochem. 1997; melanomas. Mod Pathol. 1990;3:494501.
45:327343. 40. Banks ER, Jansen JF, Oberle E, et al. Cytokeratin positivity in fine-
19. Iyer PV, Leong AS. Poorly differentiated squamous cell carcinomas needle aspirates of melanomas and sarcomas [see comments]. Diagn
of the skin can express vimentin. J Cutan Pathol. 1992;19:3439. Cytopathol. 1995;12:230233.
20. Domagala W, Lasota J, Bartkowiak J, et al. Vimentin is 41. Battifora H. Misuse of the term expression [letter; comment]. Am
preferentially expressed in human breast carcinomas with low J Clin Pathol. 1989;92:708709.
estrogen receptor and high Ki-67 growth fraction. Am J Pathol. 42. Swanson PE. Heffalumps, jagulars, and cheshire cats. A commen-
1990;136:219227. tary on cytokeratins and soft tissue sarcomas [see comments]. Am
21. Gustafsson H, Virtanen I, Thornell LE. Expression of cytokeratins J Clin Pathol. 1991;95:S2S7.
and vimentin in salivary gland carcinomas as revealed with 43. Swanson PE. HIERanarchy: the state of the art in immunohisto-
monoclonal antibodies. Virchows Arch A Pathol Anat Histopathol. chemistry [editorial]. Am J Clin Pathol. 1997;107:139140.
1988;412:515524. 44. Bacchi CE, Zarbo RJ, Jiang JJ, et al. Do glioma cells express
22. Dabbs DJ, Geisinger KR, Norris HT. Intermediate filaments in cytokeratin? Appl Immunohistochem. 1995;3:4553.
endometrial and endocervical carcinomas. The diagnostic utility of 45. Harvey JM, Clark GM, Osborne CK, et al. Estrogen receptor status
vimentin patterns. Am J Surg Pathol. 1986;10:568576. by immunohistochemistry is superior to the ligand-binding assay for

r 2006 Lippincott Williams & Wilkins 245


Yaziji and Barry Adv Anat Pathol ! Volume 13, Number 5, September 2006

predicting response to adjuvant endocrine therapy in breast cancer. 54. Yao X, Teruya-Feldstein J, Raffeld M, et al. Peripheral T-cell
J Clin Oncol. 1999;17:14741481. lymphoma with aberrant expression of CD79a and CD20:
46. Bardou VJ, Arpino G, Elledge RM, et al. Progesterone receptor a diagnostic pitfall. Mod Pathol. 2001;14:105110.
status significantly improves outcome prediction over estrogen 55. Lai R, Weiss LM, Chang KL, et al. Frequency of CD43 expression
receptor status alone for adjuvant endocrine therapy in two large in non-Hodgkin lymphoma. A survey of 742 cases and further
breast cancer databases. J Clin Oncol. 2003;21:19731979. characterization of rare CD43+ follicular lymphomas. Am J Clin
47. Bonds LA, Barnes P, Foucar K, et al. Acetic Acid-zinc-formalin: Pathol. 1999;111:488494.
a safe alternative to B-5 fixative. Am J Clin Pathol. 2005;124: 56. Marshall-Taylor CE, Cartun RW, Mandich D, et al. Immuno-
205211. histochemical detection of immunoglobulin light chain expression in
48. Matthews JB, Mason GI. Influence of decalcifying agents on B-cell non-Hodgkin lymphomas using formalin-fixed, paraffin-
immunoreactivity of formalin-fixed, paraffin-embedded tissue. embedded tissues and a heat-induced epitope retrieval technique.
Histochem J. 1984;16:771787. Appl Immunohistochem Mol Morphol. 2002;10:258262.
49. Athanasou NA, Quinn J, Heryet A, et al. Effect of decalcification 57. Du MQ, Liu H, Diss TC, et al. Kaposi sarcoma-associated
agents on immunoreactivity of cellular antigens. J Clin Pathol. herpesvirus infects monotypic (IgM lambda) but polyclonal naive
1987;40:874878. B cells in Castleman disease and associated lymphoproliferative
50. Almasri NM, Duque RE, Iturraspe J, et al. Reduced expression of disorders. Blood. 2001;97:21302136.
CD20 antigen as a characteristic marker for chronic lymphocytic 58. Attygalle AD, Liu H, Shirali S, et al. Atypical marginal zone
leukemia. Am J Hematol. 1992;40:259263. hyperplasia of mucosa-associated lymphoid tissue: a reactive
51. Wiestner A, Rosenwald A, Barry TS, et al. ZAP-70 expression condition of childhood showing immunoglobulin lambda light-
identifies a chronic lymphocytic leukemia subtype with unmutated chain restriction. Blood. 2004;104:33433348.
immunoglobulin genes, inferior clinical outcome, and distinct gene 59. Brousset P, Butet V, Chittal S, et al. Comparison of in situ
expression profile. Blood. 2003;101:49444951. hybridization using different nonisotopic probes for detection of
52. Cragg MS, Bayne MC, Illidge TM, et al. Apparent modulation of Epstein-Barr virus in nasopharyngeal carcinoma and immunohisto-
CD20 by rituximab: an alternative explanation. Blood. 2004; chemical correlation with anti-latent membrane protein antibody.
103:39893990. Author reply 39903991. Lab Invest. 1992;67:457464.
53. Barry TS, Jaffe ES, Sorbara L, et al. Peripheral T-cell lymphomas 60. Murono S, Yoshizaki T, Tanaka S, et al. Detection of Epstein-Barr
expressing CD30 and CD15. Am J Surg Pathol. 2003;27: virus in nasopharyngeal carcinoma by in situ hybridization and
15131522. polymerase chain reaction. Laryngoscope. 1997;107:523526.

246 r 2006 Lippincott Williams & Wilkins

View publication stats

You might also like